Roboflow Dataset Integration Guide

Mission Tomorrow Career Exploration Event hosted by Chamber RVA
Presented to 11,000+ eighth graders in Richmond
Volunteered for IEEE Region3 Richmond

Overview

This guide explains how to locate datasets for emotion detection from Roboflow Universe and configure the dataset download script. This is Step 1 of the IEEE Mission Tomorrow emotion detection workflow.

Roboflow provides a platform for hosting, managing, and distributing computer vision datasets. The dataset-download-roboflow.py script uses the Roboflow API to automatically download emotion detection datasets in YOLOv11 format.

Why Datasets Matter in AI

Before machines can detect emotions, they need to learn from examples. That's where datasets come in:

Dataset = Thousands of labeled images showing different emotions
Training = Teaching an AI model to recognize patterns
Deployment = Using the trained model in real applications

Think of it like teaching a friend: "This is what 'happy' looks like, this is 'sad', this is 'angry'..." After seeing hundreds of examples, your friend gets good at recognizing emotions!

Step 1: Finding Datasets in Roboflow Universe

What is Roboflow Universe?

Roboflow Universe is a public repository of datasets where users share their projects. You can access it at:

https://universe.roboflow.com/

Finding Emotion Detection Datasets

Visit Roboflow Universe: https://universe.roboflow.com/
Search for emotion datasets:
- Search terms: "emotion detection", "facial emotion", "expression recognition"
- Filter by: Dataset → Computer Vision Task → Object Detection
Example Public Datasets:
- Facial Emotion Dataset (various versions)
- Face Expression Recognition Dataset
- Emotion Classification Dataset

Dataset URL Format

When you find a dataset, the URL will look like:

https://universe.roboflow.com/WORKSPACE_NAME/PROJECT_NAME/dataset/VERSION_NUMBER

Components:

WORKSPACE_NAME — The owner's workspace identifier
PROJECT_NAME — The dataset project name
VERSION_NUMBER — The specific version of the dataset

Example:

https://universe.roboflow.com/sina-awwg4/face-emotion-detection-fanof/dataset/5

Breaking this down:

Workspace: sina-awwg4
Project: face-emotion-detection-fanof
Version: 5

Step 2: Getting Your API Key

Option A: Using a Public Dataset (No API Key Needed)

For publicly available datasets in Roboflow Universe, you can sometimes download without authentication. However, using an API key allows:

Automatic downloading
Version tracking
Workspace access

Option B: Creating Your Own Dataset

If you want to create a custom dataset:

Create a Roboflow Account:
- Go to https://roboflow.com
- Sign up (free tier available)
Create a Workspace:
- Create a new workspace for your project
- Upload your annotated images
- Configure preprocessing and augmentation
Get Your API Key:
- Click on your profile icon (top-right)
- Select "Account"
- Copy your Private API Key
- Keep this secret — never commit to public repositories

Step 3: Extracting Dataset Information

Finding Workspace and Project Names

Method 1: From the URL

https://universe.roboflow.com/sina-awwg4/face-emotion-detection-fanof/dataset/5
                           ↓              ↓
                       WORKSPACE      PROJECT

Method 2: From Roboflow Dashboard

Log in to https://roboflow.com
Navigate to your project
Go to Export/Download section
The code snippet will show your workspace and project names

Finding Version Number

The version is shown in the URL or in the project's Versions tab
Latest version is recommended
Each version may have different preprocessing applied

Step 4: Configuring the Script

Edit dataset-download-roboflow.py:

from roboflow import Roboflow

# Replace with your API key (from https://roboflow.com/settings/account)
rf = Roboflow(api_key="YOUR_API_KEY_HERE")

# Replace with your workspace name
# Found in: https://universe.roboflow.com/YOUR_WORKSPACE_NAME/...
project = rf.workspace("your_workspace_name").project("your_project_name")

# Replace with desired version number
version = project.version(5)

# Download in YOLOv11 format
dataset = version.download("yolov11")

Step 5: Example Configurations

Example 1: Using the Public Emotion Dataset

from roboflow import Roboflow

# API key (required for downloading)
rf = Roboflow(api_key="YOUR_API_KEY")

# Public dataset in Roboflow Universe
project = rf.workspace("sina-awwg4").project("face-emotion-detection-fanof")
version = project.version(5)
dataset = version.download("yolov11")

Dataset URL: https://universe.roboflow.com/sina-awwg4/face-emotion-detection-fanof/dataset/5

Example 2: Your Custom Dataset

from roboflow import Roboflow

# Your API key from account settings
rf = Roboflow(api_key="f2hJNi4Y8VSJSVHT8a66")

# Your workspace and project
project = rf.workspace("your-workspace").project("your-emotion-dataset")
version = project.version(3)
dataset = version.download("yolov11")

Step 6: Running the Script

Prerequisites

Install Roboflow Python SDK:

pip install -r roboflow-requirements.txt
# OR manually:
pip install roboflow

Download Dataset

python dataset-download-roboflow.py

Output:

Downloads to ./datasets/ (or current directory)

Creates YOLOv11-compatible structure:

datasets/
├── data.yaml           # Dataset configuration
├── train/
│   ├── images/        # Training images
│   └── labels/        # Training annotations
├── valid/
│   ├── images/
│   └── labels/
└── test/
    ├── images/
    └── labels/

Understanding Dataset Structure

data.yaml

This file is used by YOLOv11 for training:

path: /path/to/datasets
train: train/images
val: valid/images
test: test/images

nc: 10  # Number of classes (emotions)
names: ['Angry', 'Disgust', 'Excited', 'Fear', 'Happy', 
        'Sad', 'Serious', 'Thinking', 'Worried', 'neutral']

Image Annotations

Each image has a corresponding .txt label file in the same directory structure:

image.jpg → image.txt

Format (YOLO):
<class_id> <x_center> <y_center> <width> <height>
<class_id> <x_center> <y_center> <width> <height>
...

Troubleshooting

Issue: "Authentication Failed"

Verify API key is correct
Check that API key is from the correct Roboflow account
Ensure API key has permission to access the project

Issue: "Project Not Found"

Verify workspace and project names match the URL
Check for typos (case-sensitive)
Confirm project is public or you have access

Issue: "No Such Format Available"

Ensure "yolov11" is the format requested (not "yolov8")
Check Roboflow documentation for available formats

Issue: Downloaded Dataset is Empty

Verify the version number is correct
Check that the project has data uploaded
Confirm sufficient disk space

Security Best Practices

Important Security Notes:

Never commit API keys to version control:

# Add to .gitignore
echo "*api*key*" >> .gitignore
echo "*secret*" >> .gitignore

Use environment variables instead:

import os
api_key = os.getenv('ROBOFLOW_API_KEY')
rf = Roboflow(api_key=api_key)

Rotate API keys regularly in Roboflow account settings
Limit API key permissions to only required projects

Next Steps

After downloading the dataset:

Prepare for training:

cd ../yolo-model-training
cp ../roboflow/datasets/* .

Train the model:

python model-training.py --model yolo11n.pt --data data.yaml

Evaluate and export:
- Check training results in runs/detect/train/
- Export best model: runs/detect/train/weights/best.pt

Additional Resources

Roboflow Website: https://roboflow.com
Roboflow Universe: https://universe.roboflow.com
Roboflow Docs: https://docs.roboflow.com
YOLOv11 Guide: https://docs.ultralytics.com/models/yolov11/
API Reference: https://docs.roboflow.com/python-sdk

Example Workflows

Workflow 1: Use Public Dataset

1. Find dataset in Roboflow Universe
2. Copy workspace/project names from URL
3. Set API key (if required)
4. Run dataset-download-roboflow.py
5. Dataset ready for training

Workflow 2: Create Custom Dataset

1. Upload annotated images to Roboflow
2. Configure preprocessing (resize, augment)
3. Export YOLOv11 format
4. Update script with workspace/project/version
5. Run dataset-download-roboflow.py
6. Train custom model

License

This project is licensed under the MIT License. See LICENSE file for full details.

MIT License Summary: You are free to use, modify, and distribute this software for any purpose, provided you include the original license and copyright notice.

Credits & Acknowledgments

Created for: IEEE Mission Tomorrow Career Exploration Event
Event: Hosted by Chamber RVA for 11,000+ eighth graders in Richmond
Presented by: IEEE Region 3 Richmond

External Dependencies:

Roboflow — Dataset hosting and distribution platform
Roboflow Python SDK — Dataset downloading and management
YOLOv11 Format — Object detection data format standard
OpenCV — Computer vision library

Data Sources:

Public datasets from Roboflow Universe

Special Thanks:

IEEE Region 3 Richmond for volunteering
Chamber RVA for organizing Mission Tomorrow
All educators supporting STEM education

Last Updated: November 2025
Compatible With: YOLOv11, Roboflow Python SDK 0.2.32+

MalarGIT2023/roboflow-dataset-manager